A5072810129- Statistical Methods and Bayesian Inference
- Advanced Causal Inference Techniques
- Statistical Methods in Clinical Trials
- Statistical Methods and Inference
- Health Systems, Economic Evaluations, Quality of Life
- Meta-analysis and systematic reviews
- Space Satellite Systems and Control
- Spacecraft Dynamics and Control
- GNSS positioning and interference
- Astro and Planetary Science
- Inertial Sensor and Navigation
- Target Tracking and Data Fusion in Sensor Networks
- Bayesian Methods and Mixture Models
- Advanced Statistical Methods and Models
- Survey Methodology and Nonresponse
- Bayesian Modeling and Causal Inference
- Health, Environment, Cognitive Aging
- Health disparities and outcomes
- Ethics in Clinical Research
- Delphi Technique in Research
- Fault Detection and Control Systems
- Primary Care and Health Outcomes
- Planetary Science and Exploration
- Economic and Environmental Valuation
- Marine animal studies overview
MRC Clinical Trials Unit at UCL
2016-2025
London School of Hygiene & Tropical Medicine
2016-2025
University College London
2016-2025
Medical Research Council
2014-2024
University of London
2015-2024
Health Data Research UK
2019-2024
University of Leeds
2024
Atacama Large Millimeter Submillimeter Array
2023
Cincinnati Children's Hospital Medical Center
2023
American Society of Safety Professionals
2023
Non-randomised studies of the effects interventions are critical to many areas healthcare evaluation, but their results may be biased. It is therefore important understand and appraise strengths weaknesses. We developed ROBINS-I ("Risk Of Bias In Studies - Interventions"), a new tool for evaluating risk bias in estimates comparative effectiveness (harm or benefit) from that did not use randomisation allocate units (individuals clusters individuals) comparison groups. The will particularly...
Most studies have some missing data. Jonathan Sterne and colleagues describe the appropriate use reporting of multiple imputation approach to dealing with them
Funnel plots, and tests for funnel plot asymmetry, have been widely used to examine bias in the results of meta-analyses. asymmetry should not be equated with publication bias, because it has a number other possible causes. This article describes how interpret recommends appropriate tests, explains implications choice meta-analysis model
Since the early 1980s, a bewildering array of methods for constructing bootstrap confidence intervals have been proposed. In this article, we address following questions. First, when should be used. Secondly, which method chosen, and thirdly, how it implemented. order to do this, review common algorithms resampling intervals, together with some less well known ones, highlighting their strengths weaknesses. We then present simulation study, flow chart choosing an appropriate survival analysis example.
The heterogeneity statistic I 2, interpreted as the percentage of variability due to between studies rather than sampling error, depends on precision, that is, size included. Based a real meta-analysis, we simulate artificially 'inflating' sample under random effects model. For given inflation factor M = 1, 3,... and for each trial i, create M-inflated by drawing treatment effect estimate from model, using /M within-trial variance. As precision increases, while estimates variance τ 2 remain...
The Kalman filter provides an effective solution to the linear Gaussian filtering problem. However where there is nonlinearity, either in model specification or observation process, other methods are required. Methods known generically as 'particle filters' considered. These include condensation algorithm and Bayesian bootstrap sampling importance resampling (SIR) filter. filters represent posterior distribution of state variables by a system particles which evolves adapts recursively new...
Loss to follow-up is often hard avoid in randomised trials. This article suggests a framework for intention treat analysis that depends on making plausible assumptions about the missing data and including all participants sensitivity analyses
Multivariate imputation by chained equations (MICE) is commonly used for imputing missing data in epidemiologic research. The “true” model may contain nonlinearities which are not included default models. Random forest a machine learning technique can accommodate and interactions does require particular regression to be specified. We compared parametric MICE with random forest-based algorithm 2 simulation studies. first study 1,000 samples of 2,000 persons drawn from the 10,128 stable angina...
This paper provides an overview of multiple imputation and current perspectives on its use in medical research. We begin with a brief review the problem handling missing data general place this context, emphasizing relevance for longitudinal clinical trials observational studies covariates. outline how proceeds practice then sketch rationale. explore obtaining proper imputations some detail distinguish two main classes approach, methods based fully multivariate models, those that iterate...
Missing covariate data commonly occur in epidemiological and clinical research, are often dealt with using multiple imputation (MI). Imputation of partially observed covariates is complicated if the substantive model non-linear (e.g. Cox proportional hazards model), or contains squared) interaction terms, standard software implementations MI may impute from models that incompatible such models. We show how by fully conditional specification, a popular approach for performing MI, can be...
In meta-analyses, it sometimes happens that smaller trials show different, often larger, treatment effects. One possible reason for such 'small study effects' is publication bias. This said to occur when the chance of a being published increased if shows stronger effect. Assuming no other small effects, under null hypothesis bias, there should be association between effect size and precision (e.g. inverse standard error) among in meta-analysis.A number tests effects/publication bias have...
Missing data are a frequently encountered problem in epidemiologic and clinical research.[1][1],[2][2] One approach is to include the analysis only those participants without missing observations (complete or available case analysis).[1][1]–[4][3] However, addition reducing statistical
Multiple imputation is increasingly recommended in epidemiology to adjust for the bias and loss of information that may occur analyses restricted study participants with complete data ("complete-case analyses"). However, little guidance available on applying method, including which variables include model number imputations needed. Here, authors used multiple analyze prevalence wheeze among 81-month-old children Avon Longitudinal Study Parents Children (Avon, United Kingdom; 1991-1999)...
Multiple imputation is becoming increasingly established as the leading practical approach to modelling partially observed data, under assumption that data are missing at random. However, many medical and social datasets multilevel, this structure should be reflected not only in model of interest, but also model. In particular, reflect differences between level 1 variables 2 (which constant across units). This led us develop <b>REALCOM-IMPUTE</b> software, which we describe article. software...
Care home residents are at particular risk from medication errors, and our objective was to determine the prevalence potential harm of prescribing, monitoring, dispensing administration errors in UK care homes, identify their causes.A prospective study a random sample within purposive homes three areas. Errors were identified by patient interview, note review, observation practice examination dispensed items. Causes understood theoretically framed interviews with staff, doctors pharmacists....
Intention-to-treat (ITT) analysis requires all randomised individuals to be included in the groups which they were randomised. However, there is confusion about how ITT should performed presence of missing outcome data.
To resolve uncertainty as to the risk of Sudden Infant Death Syndrome (SIDS) associated with sleeping in bed your baby if neither parent smokes and is breastfed.Bed sharing was defined a parents' bed; room room. Frequency during last sleep compared between babies who died SIDS living control infants. Five large case-control datasets were combined. Missing data imputed. Random effects logistic regression controlled for confounding factors.Home arrangements infants 19 studies across UK, Europe...
Protocol deviations, for example, due to early withdrawal and noncompliance, are unavoidable in clinical trials. Such deviations often result missing data. Additional assumptions then needed the analysis, these cannot be definitively verified from data at hand. Thus, as recognized by recent regulatory guidelines reports, clarity about their implications is vital both primary analysis framing relevant sensitivity analysis. This article focuses on trials with longitudinal quantitative outcome...
Estimation of the effect a binary exposure on an outcome in presence confounding is often carried out via regression modelling. An alternative approach to use propensity score methodology. The conditional probability receiving given observed covariates and can be used, under assumption no unmeasured confounders, estimate causal exposure. In this article, we provide non-technical intuitive discussion methodology, motivating by analogy with randomised studies, describe four main ways which...